09:40
Session 6: Topology Optimization
Chair: David Lowther
09:40
20 mins
|
DESIGN OF OPTIMAL MAGNETIC CIRCUITS BASED ON EFFICIENT PENALIZATION FUNCTIONS VIA DENSITY METHODS
Satafa Sanogo, Frédéric Messine
Abstract: The design of magnetic circuit structures for electric space thrusters is difficult inverse problems. Indeed, they are often formulated as discrete nonlinear and nonconvex large-scale optimization ones. To solve them the variables has to be relaxed from discrete to continuous and then by using an adjoint variable method to compute the gradient, a local search
optimization algorithm can be performed. Nevertheless, a penalization method has to be added to render the optimal solutions the closest possible as discrete ones. In this paper, we focus on this last point by proposing some generalized interpolation schemes and by showing their great efficiency on numerical examples.
|
10:00
20 mins
|
AN ITERATIVE METHOD FOR SELECTING DECISION VARIABLES IN ANALYTICAL OPTIMIZATION PROBLEMS
Houdhayfa Ounis, Xavier Roboam, Bruno Sareni
Abstract: In this paper, we present an iterative method to assist the designer in the setting of decision variables in an optimization problem with analytical models. This method is based on a Direct Structure Matrix (DSM) which allows a clear representation of interactions between variables. This approach is applied to the geometrical model of a High Speed Permanent Magnet Synchronous Machine (HSPMSM).
|
|
